Longitudinal Monitoring of Biofilm Formation via Robust Surface

Mar 28, 2018 - (49) The UV–vis data is also plotted in Figure 4c for comparison with SERRS data. The LOD of SERRS observed in Figure 4b is 1 ng·mLâ...
0 downloads 8 Views 1MB Size
Subscriber access provided by UNIVERSITY OF TOLEDO LIBRARIES

Biological and Medical Applications of Materials and Interfaces

Longitudinal Monitoring of Biofilm Formation via Robust SERS Quantification of Pseudomonas aeruginosa-Produced Metabolite Cuong Quoc Nguyen, William John Thrift, Arunima Bhattacharjee, Saba Ranjbar, Tara Gallagher, Mahsa Darvishzadeh-Varcheie, Robert Noboru Sanderson, Filippo Capolino, Katrine Whiteson, Pierre Baldi, Allon I. Hochbaum, and Regina Ragan ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.7b18592 • Publication Date (Web): 28 Mar 2018 Downloaded from http://pubs.acs.org on March 28, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Longitudinal Monitoring of Biofilm Formation via Robust SERS Quantification of Pseudomonas aeruginosa-Produced Metabolite Cuong Quoc Nguyen,a William John Thrift,a Arunima Bhattacharjee,a Saba Ranjbar,a Tara Gallagher,b Mahsa Darvishzadeh-Varcheie,c Robert Noboru Sanderson,d Filippo Capolino,c Katrine Whiteson,b Pierre Baldi,e Allon I. Hochbaum,a and Regina Ragan a* a

Department of Chemical Engineering and Materials Science, University of California, Irvine,

Irvine, California 92697, United States; bDepartment of Molecular Biology and Biochemistry, University of California, Irvine, Irvine, California 92697, United States; cDepartment of Electrical Engineering and Computer Science, University of California, Irvine, Irvine, California 92697, United States; dDepartment of Physics and Astronomy, University of California, Irvine, Irvine, California 92697, United States; eDepartment of Computer Science, University of California, Irvine, Irvine, CA 92617, United States *Corresponding author; Email: [email protected] KEYWORDS: self-assembly, surface enhanced Raman scattering, statistical spectral analysis, biofilms, metabolomics, quorum sensing, biosensor

ACS Paragon Plus Environment

1

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 38

ABSTRACT Detection of bacterial metabolites at low concentrations in fluids with complex background allows for applications ranging from detecting biomarkers of respiratory infections to identifying contaminated medical instruments. Surface enhanced Raman scattering (SERS) spectroscopy, when utilizing plasmonic nanogaps, has the relatively unique capacity to reach trace molecular detection limits in a label free format, yet large-area device fabrication incorporating nanogaps with this level of performance has proven difficult.

Here we

demonstrate advantages of using chemical assembly to fabricate SERS surfaces with controlled nanometer gap spacings between plasmonic nanospheres. Control of nanogap spacings via the length of the chemical crosslinker provides uniform SERS signals, exhibiting detection of pyocyanin, a secondary metabolite of Pseudomonas aeruginosa, in aqueous media at concentration of 100 pg·mL-1. When using machine learning algorithms to analyze SERS data of conditioned medium from a bacterial culture, having a more complex background, we achieve 1 ng·mL-1 limit of detection of pyocyanin and robust quantification of concentration spanning fiveorders of magnitude. Nanogaps are also incorporated in an in-line microfluidic device, enabling longitudinal monitoring of P. aeruginosa biofilm formation via rapid pyocyanin detection in medium effluent as early as 3 hours after inoculation and quantification in under 9 hours. Surface-attached bacteria exposed to a bactericidal antibiotic were differentially less susceptible after 10 h of growth, indicating that these devices may be useful for early intervention of bacterial infections.

ACS Paragon Plus Environment

2

Page 3 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

INTRODUCTION During biofilm formation, differential gene expression is regulated through a cell densitydependent mechanism called quorum sensing (QS).1,2 Soon after surface attachment, bacteria begin producing extracellular polymeric substances (EPS) and QS signaling molecules.3 Once formed, a combination of physical mechanisms and genetic and metabolic adaptations within biofilms impart extreme antibiotic tolerance or resistance to constituent cells,4 which can withstand up to 1000 times higher doses of antibiotics than their free floating planktonic counterparts.5 While new antimicrobial strategies are being developed to combat antibiotic resistance, here we investigate a promising parallel strategy, sensing bacterial metabolites associated with QS for early detection of Pseudomonas aeruginosa biofilm formation at a stage where antibiotic treatment has higher efficacy. P. aeruginosa is a biofilm-forming, opportunistic pathogen that is associated with contamination of medical devices, respiratory infections in immunocompromised patients,6,7 and is one of the most common bacteria isolated in chronic wounds.8 Among the many virulence factors and QS compounds that P. aeruginosa produces is pyocyanin,9 a redox-active secondary metabolite which can act as a terminal signaling factor in the QS process. Interestingly, recent Raman spectroscopy-based studies have demonstrated the ability to distinguish between species of bacteria,10–12 identify molecules enriched in particular species, and even detect specific local metabolic activity in environmental samples using stable isotope probing techniques.13 Raman spectroscopy has been demonstrated to outperform matrix-assisted laser desorption ionization–time of flight mass spectrometry in accurately discriminating between Acinetobacter baumannii strains.12 Raman spectra provide “molecular fingerprints” composed of the vibrational spectrum of molecules serving as a label free detection method.

ACS Paragon Plus Environment

3

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 38

Biologically relevant concentrations of metabolites often range from nM to mM levels in clinical samples,14,15 and the ability to detect them has enormous potential for enabling personalized medicine. The limit of detection and quantification range of metabolite concentration needed at physiologically significant levels in biological samples, though this may vary with molecules and environments,16 is often in the µM range,17 and thus requires large and reproducible enhancements of Raman signals. Surface enhanced Raman scattering (SERS), when employing plasmonic nanogaps, is capable of providing necessary enhancements to achieve detection at biologically relevant concentrations. SERS has been used to distinguish between colony biofilms of bacteria species during maturation phases18 and even detect pyocyanin in spiked saliva down to 2.1 µg·mL-1 (10 µM)19 and spiked in subcutaneous implants in mice down to 0.1 µM.20 Yet challenges still exist in incorporating nanogaps with large and uniform enhancement factors in device architectures with nM detection limits and quantification spanning the range of metabolite concentrations found in biological samples. SERS enhancement factors due to plasmonic nanogaps are highly dependent on the distance between plasmonic nanoantennas, increasing monotonically with decreasing gap size. Statistical analysis of various size controlled nanogaps using DNA-tethering observed single molecule SERS intensity when nanogaps are on the order of 0.5 – 0.9 nm.21,22 At nanogap distances below approximately 0.5 nm, depolarization effects attributed to quantum tunneling reduce enhancements.23 Reaching sub-nanometer nanogap dimensions over large area without large variations is difficult and thus, SERS substrates often exhibit tradeoffs between reproducibility and large enhancement factors. Yet it is necessary to have both uniform and large enhancement factors across SERS substrates to reproducibly achieve low detection limits in quantitative sensing applications. Consider that at extremely low concentrations, analyte molecules will not

ACS Paragon Plus Environment

4

Page 5 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

be uniformly distributed across the surface.24,25 SERS measurements on a mixture of two different analytes determined that single molecule sensing events occur at nM concentration as not every molecule in the scattering volume will reside in a hotspot.25,26 Inherent variances in SERS substrates’ enhancement factors only worsen any analyte’s location-dependent signal variations and will lead to large SERS intensity fluctuations. While such surfaces may produce a low limit of detection due to the presence of a few hotspots on the surface, the limit of quantification will be much higher. Thus increasing the uniformity and density of nanogaps with high enhancements will lead to a higher probability that an analyte will adsorb on a hotspot in the illuminating laser spot size and thereby reproducibly contribute to the SERS signal at low concentrations. Here, large scale uniformity in hot spot intensity from nanoantennas on SERS substrates is achieved using a hierarchical self-assembly method for fabricating Au nanospheres into closepacked structures, referred to as oligomers, on a diblock copolymer template.

Assembly

combines long range, electrohydrodynamic (EHD) flow,27–30 and short range,

chemical

crosslinking, driving forces to provide a high density of oligomers with sub-nanometer nanogap spacing. Induced by an electric field in colloidal solution, EHD flow promotes lateral motion and close-packing of nanospheres in colloid at the electrode surfaces, assembling transient closepacked oligomers. Transient oligomers are frozen using EDC, a carbodiimide crosslinker, to form an anhydride bridge between the carboxylic acid ligands on Au nanospheres. We have previously reported that this process achieves a high yield of close-packed oligomers composed of ten nanospheres or less with gap spacing of 0.9 nm, having an electric field enhancement on the order of 600 in the hotspot region.31 Close-packing is also advantageous as it allows for variability of oligomer orientation with respect to polarization of incident light. In a close

ACS Paragon Plus Environment

5

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 38

packed oligomer, the plasmon resonance will most closely reflect that of a linear oligomer along the polarization axis.31,32 When using benzenethiol as the analyte, here we report that SERS signal exhibits a relative standard deviation of 10% across a 100 µm × 100 µm area with over 1000 different measurements. This achievable uniform SERS response allows for incorporation in microfluidic device architectures and spectral data analysis using multivariate machine learning algorithms.33– 35

A large number of training data sets and the full spectra collected from SERS substrates are

necessary for accurate quantitative analysis by accounting for signal variance inherent to SERS. By combining uniform SERS substrates having high signal enhancements with multivariate statistical analysis of SERS spectra, we are able to differentiate pyocyanin in the complex soup of biological media at concentrations down to 1 ng·mL-1 (4.8 nM) and robustly quantify concentrations spanning five-orders of magnitude. Moreover rapid SERS analysis in solutions negates the need to grow biofilm directly on SERS substrates20 or to drop cast plasmonic clusters on static biofilms,18 thus allowing instantaneous quantification of pyocyanin for detection of biofilm formation on any surface. Longitudinal monitoring of supernatant from bacterial cultures in microfluidic devices exhibits SERS signal from pyocyanin as early as 3 h after P. aeruginosa culture inoculation. This is before surface-attached bacteria exhibits a decreased susceptibility to bactericidal antibiotic carbenicillin that was measured at 10 h. Low cost fabrication of nanoantennas with nanometer scale nanogaps over large areas enabled by self-assembly and statistical spectral analysis of SERS data demonstrate the capacity for fabricating device architectures capable of early detection of biofilms enabling effective antimicrobial treatment. RESULTS AND DISCUSSION

ACS Paragon Plus Environment

6

Page 7 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Large-Area Uniformity of SERS Substrates Directing assembly of Au oligomers with uniformly distributed electromagnetic hotspots over large areas is achieved by utilizing electrohydrodynamic (EHD) flow to drive chemical crosslinking to form nanogaps with sub-nanometer gap spacing as reported previously.31 Chemical assembly is facilitated using a two-step growth process: (1) electrophoretic sedimentation driving chemical crosslinking of Au nanoparticle seeds onto a working electrode and (2) growing oligomers via EHD flow, and stabilizing via chemical crosslinking. During electrophoresis, depicted in Figure 1a, lipoic acid functionalized Au nanoparticle monomer seeds are chemically assembled on amine functionalized PMMA regions of a diblock copolymer poly(styrene-b-methyl methacrylate) (PS-b-PMMA) coated working electrode using 1-Ethyl-3(3-(dimethylamino) propyl)carbodiimide/N-hydroxysulfosuccinimide (EDC/sNHS) carbodiimide chemistry (inset). In the second deposition step, an applied field deforms the ionic double layer around the Au seeds, shown on the right of Figure 1a, inducing an osmotic flow towards the seeds’ equator, termed EHD flow.30,36,37 These flow fields entrain nearby nanospheres and drives them toward Au seeds to form transient close packed oligomers, which are subsequently stabilized through an acylisourea – carboxylic acid reaction that forms anhydride bridges between nanospheres, shown in Figure 1b. These bridges result in oligomers with uniform interparticle gap spacings of approximately 0.9 nm, corresponding to the sulfur-sulfur distance of an anhydride bridge calculated from atomistic simulations and observed in TEM.31 A scanning electron microscopy (SEM) image of a SERS substrate composed of self-assembled oligomers is displayed in Figure 2a. The size distribution in percent area (Figure 2b inset), determined from image analysis of SEM images using Wolfram MathematicaTM, indicates that trimers are observed with the highest probability, with quadrumers being the second. The

ACS Paragon Plus Environment

7

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 38

absorption cross section of a dimer, linear trimer, and linear quadrumer with a nanogap size of 0.9 nm – corresponding to the calculated sulfur-sulfur distance of an anhydride bridge – was calculated from full-wave simulations and shows good agreement with the fine structure observed in UV-visible measurement as shown in Figure 2b. This agreement suggests that measured attenuation is dominated by frequencies near the trimer resonance with contributions from frequencies associated with dimers and quadrumers, blue and red shifted, respectively. Larger oligomers observed in the SEM image of Figure 2a contribute to the broad shoulder at higher wavelength. This observation can be understood by considering the effect of the close packed arrangement of nanospheres in oligomers on their plasmon resonance. It has been previously shown using dark field microscopy and full-wave simulations that an oligomer’s plasmon mode is only slightly perturbed by the addition of a nanosphere when the added nanosphere is unaligned with the incident beam’s polarization. For example, a hexamer made up of two trimers on top of each other has a plasmon mode that is slightly redshifted from that of a linear trimer.32 Consider that for a close packed oligomer to have more than 4 nanospheres in a row, it must be composed of more than 21 nanospheres. From the oligomer size distribution, Figure 2b inset, it was determined that, by number, 98% (88% by area) of oligomers are composed of 21 nanospheres or less. It is then unsurprising that the resonance is dominated by dimers, trimers and quadrumers. Full-wave simulations of the electric field enhancement – |Eolig|/|E0|, where Eolig and E0 are the plane wave field with and without oligomers, respectively – was performed for nanogaps of 0.9 nm and 2 nm in dimers and linear trimers and are shown in Figure 2c. As expected the resonance red shifts as the gap distance decreases. Furthermore, increasing gap spacing from 0.9 to 2 nm reduces the calculated field enhancement from 621 to 240 (762 to 314) in a dimer (linear trimer). The SERS enhancement can be estimated as the

ACS Paragon Plus Environment

8

Page 9 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

fourth power of the field enhancement, |Eolig|/|E0|4, and thus even slight variations in field enhancement will lead to large variations in SERS intensity. Overall the plasmon resonance is less affected by close packed oligomer size than the resultant field enhancement in the nanogaps when spacing decreases below 2 nm. Thus chemically controlled gap spacing enables uniform SERS intensity, observed in Figure 2d. Figure 2d displays the normalized SERS intensity of a benzenethiol vibration band, 1573 cm-1, acquired over a 100 µm × 100 µm area. The SERS intensity has a relative standard deviation (RSD) of 10.4%. While nanogap spacing of 0.9 nm is important to achieve large and uniform SERS enhancements, at the same time, the assembly method provides a sufficiently broad band response, ranging over a window of 625 nm to 875 nm in the attenuation measurement, to enhance signals at both the plasmon excitation wavelength and Raman scattered wavelength. The uniform SERS response with large enhancements over large area enables the use of these SERS substrates in device architectures. Of further significance, it also enables the acquisition of large datasets needed for statistical analysis enabling quantitative detection. Quantification and Detection of Pyocyanin in Aqueous Media Pyocyanin exhibits a broad absorption band from 550 nm to 900 nm,20,38 thus using a 785 nm laser to excite nanoantennas results in surface enhanced resonance Raman scattering (SERRS). SERRS spectra of aqueous pyocyanin from 1 µg·mL-1 (4.8 µM) to 100 µg·mL-1 (480 µM) (Figure 3a) displays clear Raman bands similar to pyocyanin spectra reported using surfaces with Ag colloids19 and Ag and Au nanorods20,39 at 552 cm-1, 1353 cm-1, 1602 cm-1, and 1620 cm-1. These bands rise at a concentration as low as of 100 pg·mL-1 (480 pM), as exemplified in Figure 3b for the band 552 cm-1. The log-log dose-response curve of pyocyanin at 552 cm-1 (Figure 3c)

ACS Paragon Plus Environment

9

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 38

reveals a linear regime between 1 ng·mL-1 and 10 µg·mL-1, consistent with Langmuir adsorption kinetics; similar linear relationships with concentration are also observed for other pyocyanin vibrational bands and results are shown in Figure S1. Within this linear regime, pyocyanin concentration can be quantified using the formula shown in equation (1) with R2 = 0.951. log C = a log I + b

(1)

Here a and b are fitting constants with value of 3.623 and 2.924, respectively, while C and I represent pyocyanin concentration and normalized SERRS intensity at 552 cm-1, respectively. For concentrations below the limit of quantification (LOQ), 1 ng·mL-1, SERS substrates detection of pyocyanin was determined by comparing signal and background at 552 cm-1 per guideline EP17 of the Clinical and Laboratory Standards Institute.40 The limit of blank (LOB) is calculated by adding the mean background signal (DI water) Ībg at 552 cm-1 to 1.645 its standard deviation σbg , shown in equation (2): LOB = Ībg + 1.645σbg

(2)

Detection is defined in the standard manner where the mean signal, ĪD, at 552 cm-1 is at least 1.645 standard deviations, σD, larger than or equal to the LOB, shown in equation (3): ĪD – 1.645σD ≥ LOB

(3)

From this analysis, SERS substrates exhibit detection of pyocyanin in aqueous media at concentration of 100 pg·mL-1, above which one can observe pyocyanin signals above the background in Figure 3c. We did not measure pyocyanin at lower concentration since our interest was in establishing a LOD in biological media. These results are discussed in the following section.

ACS Paragon Plus Environment

10

Page 11 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Training Data Acquisition and Building Multivariate Predictive Model Using linear regression on a single Raman band allows for facile comparison with other SERS surfaces in the literature as it is widely used for quantitative calibration.19,20,39,41–46 While this method is sufficient for analysis of a pure analyte in water, it discards the remaining rich spectral information in each spectrum and can lose sensitivity when other molecules are present in solution as is the case when monitoring biofilm growth. Alternatively multivariate analysis, specifically partial least square (PLS) regression, analyzes the full spectra collected from SERS substrates and improves limit of quantifications in complex media.47 Thus we capitalize on the uniform SERS response of our surfaces to acquire necessary training data and employ PLS analysis to quantify pyocyanin in more complex biological media, an important requirement for diagnostic applications. Training data sets, composed of SERRS spectra from known concentrations of pyocyanin spiked in LB media, generate a robust PLS predictive model for pyocyanin concentration in the increased background noise of the biological media. The multivariate model predictive capability is subsequently demonstrated by quantifying pyocyanin production from P. aeruginosa planktonic cultures during in vitro growth. First we demonstrate the suitability of our substrates for detecting pyocyanin in complex media by comparing SERRS spectra of blank LB media, LB media spiked with 10 µg·mL-1 of pyocyanin alongside spectra from of cell-free supernatants collected from mid-stationary phase cultures of wild type P. aeruginosa PA14 and its phenazine-deficient mutant strain ∆phz1/2 as a control (Figure 4a). The ∆phz1/2 strain does not produce phenazines, hence its SERRS spectrum

ACS Paragon Plus Environment

11

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 38

(multiplied a factor of 5 for clarity) indeed lacks the vibrational fingerprint of pyocyanin; it is similar to the spectrum obtained for blank LB media. Meanwhile, SERRS spectra of wild type PA14 and pyocyanin in LB broth exhibit similar features, including distinct pyocyanin vibrational bands, and thus confirming the suitability of our substrates for the task. While pyocyanin signals are clearly seen here, signal interference from other molecules is amplified at low concentration, requiring more the more sophisticated analysis. The calibration datasets that were acquired by collecting SERRS spectra of LB media spiked with 100 pg·mL-1 to 100 µg·mL-1 pyocyanin generated a predictive model of pyocyanin concentration in LB media using PLS regression. From the 400 spectra collected for each pyocyanin dose, 380 were randomly selected as training set and the remaining 20 withheld as testing set. To optimize the model, over- and under-fitting are avoided by using 10 PLS components where a minimum in RMSE of cross validation (RMSECV) is observed, displayed in inset of Figure 4b. With the testing set, the model demonstrates accurate prediction between 1 ng·mL-1 and 100 µg·mL-1 as shown in Figure 4b. Fitting the predicted versus actual concentration with a line having a slope of 1 – representing perfect predictive capability – gives a R2 value of 0.956. Pyocyanin Quantitative Detection in Complex Media The generated predictive model was used to quantify pyocyanin from SERRS spectra of cell-free conditioned growth medium of wild type P. aeruginosa and ∆phz1/2 as a function of incubation time. For each incubation time point, 400 SERRS spectra were collected within the span of less than 4 minutes, and their corresponding pyocyanin concentrations were calculated using the PLS model; the average concentration for incubation times ranging from 0 to 24 h are shown in

ACS Paragon Plus Environment

12

Page 13 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Figure 4c. The performance of SERRS-PLS was compared with UV-vis absorption spectroscopy typically employed to quantify pyocyanin. Pyocyanin concentrations were calculated from UVvis absorption peaks using the reported molar absorptivity  = 4.31 × 10    at  = 690  .48 The pyocyanin-deficient ∆phz1/2 strain were used to calculate the instrument noise and contributions from the background media. This yielded the UV-vis LOD as 197 ng·mL-1 (0.94 µM), equivalent to 3 standard deviations above the background of the ∆phz1/2 strain. This value agrees well with the reported LOD of 1 µM for UV-vis.49 The UV-vis data is also plotted in Figure 4c for comparison with SERRS data. The LOD of SERRS observed in Figure 4b is 1 ng·mL-1 (4.8 nM) using the same definition for differentiating from background noise and PLS model discussed above. Thus SERS substrates are able to detect pyocyanin as early as 2 h of shaking culture growth and quantify the concentration as 2.5 ng·mL-1 (12 nM), as opposed to after 8 h of shaking culture growth when using UV-vis absorption. From 8 h onward, quantitative performance of the two methods is comparable. The sharp increase in pyocyanin concentration detected by SERRS from 2 to 8 h correlates with the exponential growth phase of P. aeruginosa in shaking culture, which is observed in Figure 4d as an increase in optical density and thereby accumulation of biomass. Overall this data validates SERS substrates robustness toward detection of pyocyanin in complex media. Monitoring Biofilm Formation via Pyocyanin Quantification Here we integrate microfluidic channels with SERS substrates to perform in-line sampling of biofilm effluent, thus enabling rapid quantitative detection of pyocyanin as a means to longitudinally monitor biofilm growth. P. aeruginosa biofilms were grown in tryptone broth

ACS Paragon Plus Environment

13

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 38

(TB) medium as previously described.50 Effluent from the biofilm growth channel is delivered to a microfluidic channel with Au oligomers assembled on the surface as illustrated in Figure 5a. Every three hours, 200 SERRS spectra are collected in the effluent with a total acquisition time of 2 minutes, thus eliminating the need for performing chloroform extraction, incubating,20 or evaporating solvent on SERS surfaces.39 To quantify pyocyanin, a suitable predictive model was generated by repeating PLS analysis on training dataset acquired using the appropriate collection parameters for TB media (as oppose to LB in planktonic cultures growth) as reported in our methods. The resulting LOD from this analysis is 10 ng·mL-1. Using the model, pyocyanin was detected in the SERRS signal above the LOD starting between 6 h - 9 h after inoculating as observed in Figure 5b. Thus the time of quantification (TOQ) falls within this range. Pyocyanin concentration after 9 h was determined to be 24 ng·mL-1 (115 nM), above the determined LOD. The half-life of pyocyanin desorption from the surface was determined to be on the order of 11 minutes from subsequent SERS measurements from samples immersed in water, exposed to 1 μg·mL-1 (4.8 μM) pyocyanin, rinsed, and re-immersed in water, shown in Figure S2 of Supporting Information. Thus, the calculated concentrations are indicative of the instantaneous production of pyocyanin in biofilm growth channel. Investigation of results generated from the model shows that some spectra at 3 and 6 h reveals Raman bands associated with pyocyanin above the background. By analyzing the individual spectra it was determined that 7.8% and 14.3% of the spectra at 3 and 6 h, respectively, show a pyocyanin concentration above the background; representative spectra are plotted in Figure 5c. This fraction increases to 70.3%, 96.7%, and 100% for the subsequent time points. This is consistent with non-uniform distribution of pyocyanin on the substrate surface at low concentrations. So while the averaged concentration of pyocyanin calculated for the set of Raman spectra at 3 and 6 h of biofilm

ACS Paragon Plus Environment

14

Page 15 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

growth lies below the LOD, a fraction of those spectra shows that pyocyanin is detected at those early time points. Thus 3 hours is set as the time of detection (TOD). The calculated concentrations of pyocyanin correlate with the time-dependent accumulation of biofilm biomass obtained from analysis of fluorescence images of the microfluidic growth channels (Figure 5d). Representative confocal fluorescence images of the growth channels show the initial stages of bacterial cell adhesion (6 h), microcolony formation (10 h), and three-dimensional growth of biofilms above the channel surface (16 h). The biomass and imaging data indicates that biofilm formation can be monitored reliably after TOQ. More significantly, it reveals that the earliest detection of biofilms, corresponding to TOD, occurs during the initial stages of bacterial cell adhesion, earlier than observed in fluorescence confocal images. Bacterial biofilms impart antibiotic resistance and tolerance on constituent cells via several distinct mechanism.4 These mechanisms are characteristic of bacterial phenotypes and properties of mature biofilms, so early treatment with antibiotics can be more effective than equivalent treatment of mature biofilms. The potential for early detection to provide a therapeutic improvement for infection outcomes was assessed by measuring the antibiotic susceptibility of surface-attached bacteria at different stages of biofilm growth. Bacteria exposed to a bactericidal antibiotic, carbenicillin, were differentially susceptible after 10 h of growth, after which their susceptibility decreased significantly with increasing growth time as they transition into a more antibiotic-tolerant state (Figure 5e); this transition is also captured as the sharp increase in pyocyanin concentration measured from SERRS in Figure 5b between 9 h and 12 h. A control biofilm was grown for 24 h and exposed to PBS instead of Carb. n.s. indicates no statistically significant difference between the antibiotic susceptibility of cells grown in flow channels for 6 and 10 h. Consequently, detection of a P. aeruginosa infection and intervention at early stages of

ACS Paragon Plus Environment

15

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 38

colonization (< 10 h growth) show potential to substantially improve the effectiveness of antibiotic treatment. CONCLUSION Chemically assembled SERS substrates, using electrohydrodynamic flow to initiate chemical crosslinking, yields uniform nanogap spacings of 0.9 nm that proves advantageous for quantitative molecular detection. Full-wave simulations of near field enhancements comparing 1 nm and 2 nm nanogaps show that variation of this parameter is critical to achieve high and reproducible SERS intensity needed to generate reliable training sets for machine learning algorithms. Large area signal uniformity, low excitation power, 14.6 µW, and short integration time, 0.1 second, allows for integration of SERS substrates in a device platform and the rapid acquisition of large data sets for statistical analysis. SERRS substrates exhibit the ability to detect pyocyanin in aqueous media at concentration of 100 pg·mL-1 when individual pyocyanin Raman bands are investigated. In more complex media, PLS analysis of spectral data enables robust quantification of pyocyanin spanning 5-orders of magnitude in biologically relevant levels between 1 ng·mL-1 and 100 µg·mL-1. Integration of microfluidic architectures with SERS substrates facilitates in-line sampling of effluent medium for longitudinal monitoring of pyocyanin concentration during P. aeruginosa biofilm formation. Measured pyocyanin concentration in effluent medium correlates with accumulated biofilm biomass obtained from confocal fluorescence images. More significantly, pyocyanin can be detected as early as 3 h after inoculation. A sharp increase in pyocyanin concentration is observed between 9 and 12 h after inoculation; this corresponds to decreased susceptibility of surface-attached bacteria to bactericidal antibiotic carbenicillin at 10 h. Thus detecting microbial production of metabolites associated with quorum sensing with chemically assembled SERS substrates and analysis using

ACS Paragon Plus Environment

16

Page 17 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

machine learning algorithms lays out a promising strategy for detection of metabolites in biological media that can be pursued in future studies for more complex systems. EXPERIMENTAL METHODS Materials Random copolymer Poly(styrene-co-methyl methacrylate)-α-Hydroxyl-ω-tempo moiety (PS-rPMMA) (Mn = 7,4 kg mol-1, 59.6% PS) and diblock copolymer poly(styrene-b-methyl methacrylate) (PS-b-PMMA) (Mn = 170-b-144 kg mol-1 ) were purchased from Polymer Source, Inc. (Dorval, Canada). Gold nanospheres with diameter of 40 nm and lipoic functionalization were purchased from Nanocomposix (San Diego, CA). Si(001) wafers with resistivity of 0.004 ohm-cm were purchased from Virginia Semiconductor (Frederickburg, VA). Hydrofluoric acid (HF) was purchased from Fisher Scientific (Pittsburg, PA). 2-(N-morpholino)ethanesulfonic acid (MES)

0.1M

buffer

was

purchased

from

Pierce

(Rockford,

IL).

1-ethyl-3-[3-

dimethylaminopropyl] carbodiimide hydrochloride, N-hydroxy sulfosccinimide (s-NHS), dimethyl sulfoxide (DMSO), ethylenediamine (ED), toluene, ethanol, isopropanol (IPA), potassium carbonate, and 52-mesh Pt gauze foil were purchased from Sigma Aldrich (St. Louis, MO). Nanopure deionized (DI) water at 18.2Mohm cm-1 was obtained from Milli-Q Millipore System. Self-Assembly of SERS Substrates Si wafers were cleaned in 10% HF to remove the native oxide. The potential of HF to cause severe injury mandates extreme caution during usage. Random copolymer PS-r-PMMA and diblock copolymer PS-b-PMMA solution in toluene (1 wt%) were spun coat on the flat Si

ACS Paragon Plus Environment

17

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 38

surface and annealed at 198oC to form thin films as described in previous work.51 PMMA regions are selectively functionalized with amine end groups by first immersing in DMSO and then in ED/DMSO solution (5% v/v) for 5 minutes each step and without rinsing in between. The substrate is then washed with IPA for 1 minute and dried under nitrogen. Lipoic acid functionalized 40 nm Au nanospheres in aqueous solution is concentrated twofold by adjusting the pH to 8 with potassium carbonate and centrifuging for 25 minutes at 1700 RCF and redispersed in DI water. 3 mL of concentrated Au nanosphere solution is added to a 10 mL beaker, followed by 35 uL of freshly prepared 20 mM s-NHS in 0.1 M MES buffer and 35 uL of freshly prepared 8 mM EDC in 0.1 M MES buffer. The solution is swirled and moved on a hotplate to be heated to 8o°C. A 1 cm x 1 cm Pt mesh and 1 cm x 1 cm Si substrate are stabilized using alligator clips and placed into the solution vertically as the anode and cathode, respectively. A DC Regulated Power Supply was used to apply a voltage of 1.2V for 10 minutes to drive electrophoretic sedimentation of Au nanospheres to the surface. EDC initiated carbodiimide crosslinking chemistry is used to covalently bind carboxylic acid functionalized Au nanospheres to the selectively amine-functionalized PMMA regions of the template via an amide bond.51 The substrate, Pt mesh, and beaker are rinsed with IPA for 1 minute and dried under nitrogen. The process is repeated on the same substrate with a freshly concentrated Au nanospheres solution, but with 30uL of EDC and s-NHS solution. Here electrohydrodynamic flow around previously attached nanospheres drives nanospheres to chemically crosslink via EDC coupling chemistry to nanospheres assembled in the first step.

Finally, the diblock

copolymer layer is etched under oxygen plasma (50W, 60s) in the PC2000 Plasma Cleaner from South Bay Technology (San Clemente, CA) to remove organic groups from nanoparticle surfaces. Assembly on indium tin oxide (ITO)-coated glass are carried out identically, with the

ACS Paragon Plus Environment

18

Page 19 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

exception of the surfaces undergoing oxygen plasma etching at 100 W for 1 min before spin coating PS-r-PMMA. Characterization Images of SERS substrates are collected with a Magellan XHR scanning electron microscope (FEI). UV-vis absorption measurements of SERS substrates are carried out using a Shimadzu UV-1700 absorption spectrometer. Absorption spectra are taken of ITO-coated glass substrates taped (away from the beam path) on to a quartz cuvette filled with DI water. P. aeruginosa cell free supernatant preparation Wild type P. aeruginosa (strain PA1452) shaking culture supernatant was used to measure pyocyanin production over time. PA14 was streaked onto lysogeny broth (LB, EMD Millipore) agar plates from frozen glycerol stocks and grown overnight at 37 °C. Shaking cultures were inoculated from single colonies on the LB plate into 5 mL of liquid LB and grown at 37 °C on an orbital shaker overnight. 50 µL of the shaking culture was diluted into 24 mL of 10 g/L tryptone media (Bacto tryptone, BD Scientific) and this subculture was grown at 37 °C on an orbital shaker. Aliquots were taken periodically from the shaking subculture for optical density measurements. To measure pyocyanin production, P. aeruginosa conditioned medium was isolated by centrifuging the 2 mL cultures at 15000 x g and passing the resulting supernatant through a 0.2 µm PES vacuum filter (Corning). The filtered supernatants were dropped onto the SERS substrates and measured within 1 hour from their collection time. Fluidic Device Fabrication and Biofilm Growth

ACS Paragon Plus Environment

19

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 38

The microfluidic device consists of two channels, one for the biofilm and the other for the SERS substrate. Microfluidic channels for the biofilm were made by bonding the plasma-activated surfaces of PDMS with a glass slide. SERS substrate and a glass slide are separated by an adhesive spacer of 100 μm thickness (3M 415) with channel design laser cut (Epilog Fusion Laser Cutter) in the adhesive layer. P. aeruginosa biofilms for in-line detection of pyocyanin were grown in the above microfluidic devices as previously described using a P. aeruginosa strain constitutively expressing yellow fluorescent protein (YFP).50 P. aeruginosa cells were seeded with no flow in the biofilm growth channel for 2 h. The fluid feed was then switched to sterile tryptone media and the outlet of the channel was connected to the SERS channel input. The 0 h SERRS measurements were collected once the in-line device in fully connected, prior to start of flow. Media was then pulled through the in-line detection device at 10 µL·h-1 and SERRS spectra were collected at the indicated time points. Identical biofilms were grown and imaged in the flow cells to measure the timedependence of biofilm growth. The biofilm volume was obtained from Volocity imaging analysis software (PerkinElmer) of confocal fluorescence images of the YFP-producing biofilms. SERRS collection parameters are described below. Antibiotic Susceptibility Measurements P. aeruginosa biofilms were grown for 10, 15, and 24 h on glass coverslips submerged in 2 mL of TB supplemented with 3 g⋅L-1 NaCl in sterile, six well tissue culture plates (Fisher Scientific). After the indicated growth time, the growth media was aspirated and the biofilms on the coverslips were rinsed once with PBS while still in the wells. The PBS rinse solution was then aspirated and replaced with 600 μg⋅mL-1 carbenicillin (minimum inhibitory concentration in P.

ACS Paragon Plus Environment

20

Page 21 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

aeruginosa strain PA14 128 μg⋅mL-1)53 in PBS or PBS only for the control. Biofilms were soaked in the antibiotic or control solutions for 3 h, rinsed again with PBS, and then stained with live/dead cell viability assay stains, propidium iodide and Syto 9 (Fisher Scientific) at 2 μM final concentration each, for 15 min. Coverslips were rinsed once more in PBS, removed from the wells and placed face down on a microscope slide for confocal fluorescence imaging. The fraction of dead cells was calculated using the biovolumes of each color channel (red and green) obtained from Volocity imaging analysis software (PerkinElmer) of confocal fluorescence images of biofilms. At 6 h biofilm growth, the washing steps described above removed all cells from the coverslips. Instead, the antibiotic susceptibility of surface-attached bacteria was measured from cells grown in microfluidic devices. After 6 h of growth, as described above for pyocyanin detection experiments, the input line was changed from TB to 600 μg⋅mL-1 carbenicillin in PBS for 3 h. The input line was then changed again to BacLight for 15 min and finally PBS for 30 min, after which cells attached to the channel surface were imaged and the dead cell fraction quantified by counting individual cells of each color channel. Spectroscopic Measurements Instrumentation and Procedure Surface enhanced Raman spectroscopy (SERS) measurements are conducted using a Renishaw InVia Raman Microscope system. A 785 nm continuous wave laser is chosen to excite near the plasmon resonance of Au nanoparticle assemblies as determined from electromagnetic simulations and ultraviolet-visible absorption measurements.31 In droplet measurements, as referred to in the main text, a 60X water immersion objective with 1.2 NA is used for illumination and collection. Approximately 150 µL of solution of interest is

ACS Paragon Plus Environment

21

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 38

transferred onto SERS substrates and the measurements are acquired with laser power and acquisition time of 7.3 µW and 0.5 second, respectively, over area specified for each case. SERS substrates are cleaned with IPA and DI water for 1 minute each and dried under nitrogen between measurements. One substrate was used per set of measurements (concentrations or time points). For in-line measurements, a 50X objective is used. Measurements are taken with laser power and acquisition time of 14.6 µW and 0.1 second, respectively. Illumination and collection were done through SERS microfluidic channel. Spectra Processing and Analysis Raman scattering spectra processing and analysis were performed off-line using MATLAB R2016b (The MathWorks Inc, Natrick, MA). Each spectrum undergoes baseline correction, smoothing with Savitzky-Golay, and normalization to the average intensity of Si second-order vibrational band, reported between 920 – 1045 cm-1,54 from the substrate.

This allows for

comparison of different samples where slight intensity variations may arise due to deviations in optical collection in the experimental setup. While this band was previously only the range between 920 – 970 cm-1 was used due to the appearance of pyocyanin ring stretching vibrational band at 975 cm-1. When preparing SERRS spectra for full-spectrum partial least squares (PLS) regression, a constant  = 1.0067 was added to the processed signals to eliminate negative values in the calibration matrix associated with variations due to noise. This is necessary as signals are then log-transformed before analysis with PLS regression. PLS regression combines characteristics of principal component analysis with multiple linear regression to predict a set of dependent variables from a large set of independent variables.55

ACS Paragon Plus Environment

22

Page 23 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Simulations Full-wave simulations (frequency domain finite elements method solver) are implemented in CST Microwave Studio (CST AG). We calculate the absorption cross sections of several nanosphere oligomers: dimer, linear trimer, and linear quadrumer. The structure is as follows. Au nanospheres of 40 nm in diameter with 0.9 nm interparticle gaps31 are partially embedded in a 40 nm PMMA layer56 with their centers 8 nm above the PMMA. The PMMA layer is directly above a 150 nm ITO layer, which is directly above a 2 µm glass layer. The upper medium surrounding nanospheres is water. The total size of the structure in x-y plane (transverse plane) is assumed to be 4 µm × 4 µm, which is approximately four times larger than the largest excitation wavelength of 900 nm, and we applied software's open boundary condition. Au permittivity was calculated from Drude model with parameters extracted from Grady.57 The relative electric permittivity used for water, PMMA, glass, and ITO are 1.77, 2.47, 2.3207, and adapted from Moerland58, respectively. The permittivity in the gap region was unknown due to anhydride bonding31, and thus was approximated by performing a parameter sweep of the gap permittivity and calculating the absorption of the dimer configuration to determine which best corresponds with the observed dimer peak at 686 nm. The resulting permittivity of 2.25 was used in simulations. Oligomers are excited with plane wave illumination at normal incidence with electric field polarization along the axis of the linear oligomers, and the absorption cross section of the structures is determined. Field enhancement was also calculated from full-wave simulations of dimer and linear trimer with 0.9 nm and 2 nm interparticle gaps using the same conditions. Field enhancement is defined in equation (4). FE = |Eolig|/|E0|

(4)

ACS Paragon Plus Environment

23

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 38

|Eolig| represents the electric field magnitude at the center of the gap in the oligomer (dimer or trimer), and |E0| is the electric field magnitude at the same location in the absence of the structure. SUPPORTING INFORMATION The Supporting Information is available free of charge on the ACS Publications website at DOI: SI contains: Dose-response relationships of SERRS intensity for additional commonly observed pyocyanin vibrational modes at 1620 cm-1, 1602 cm-1, and 1353 cm-1; Details on estimation of pyocyanin desorption half-life from SERRS intensity at 1620 cm-1, 1602 cm-1, 1353 cm-1, and 552 cm-1 before, during, and after exposure to 1 μg·mL-1 (4.8 μM) pyocyanin.

ACKNOWLEDGMENT The authors acknowledge the National Science Foundation EECS-1449397 for funding this work. The authors would like to thank Computer Simulation Technology (CST) of America, Inc. for providing CST Microwave Studio. W.J.T. acknowledges support from the National Science Foundation BEST IGERT Program (NSF DGE-1144901). The authors also acknowledge the use of the facilities within the Laser Spectroscopy Facility, the Laboratory for Fluorescence Dynamics, and the Laboratory for Electron and X-ray Instrumentation (LEXI) center at the University of California, Irvine. REFERENCES (1)

Kievit, T. R. D.; Gillis, R.; Marx, S.; Brown, C.; Iglewski, B. H. Quorum-Sensing Genes in Pseudomonas Aeruginosa Biofilms: Their Role and Expression Patterns. Appl. Environ. Microbiol. 2001, 67, 1865–1873.

ACS Paragon Plus Environment

24

Page 25 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

(2)

Miller, M. B.; Bassler, B. L. Quorum Sensing in Bacteria. Annu. Rev. Microbiol. 2001, 55 (1), 165–199.

(3)

Kolter, R.; Greenberg, E. P. Microbial Sciences: The Superficial Life of Microbes. Nature 2006, 441, 300–302.

(4)

Anderson, G. G.; O’Toole, G. A. Innate and Induced Resistance Mechanisms of Bacterial Biofilms. In Bacterial Biofilms; Current Topics in Microbiology and Immunology; Springer, Berlin, Heidelberg, 2008; pp 85–105.

(5)

Malone, M.; Goeres, D. M.; Gosbell, I.; Vickery, K.; Jensen, S.; Stoodley, P. Approaches to Biofilm-Associated Infections: The Need for Standardized and Relevant Biofilm Methods for Clinical Applications. Expert Rev. Anti Infect. Ther. 2017, 15, 147–156.

(6)

Hunter, R. C.; Klepac-Ceraj, V.; Lorenzi, M. M.; Grotzinger, H.; Martin, T. R.; Newman, D. K. Phenazine Content in the Cystic Fibrosis Respiratory Tract Negatively Correlates with Lung Function and Microbial Complexity. Am. J. Respir. Cell Mol. Biol. 2012, 47, 738–745.

(7)

Whiteson, K. L.; Meinardi, S.; Lim, Y. W.; Schmieder, R.; Maughan, H.; Quinn, R.; Blake, D. R.; Conrad, D.; Rohwer, F. Breath Gas Metabolites and Bacterial Metagenomes from Cystic Fibrosis Airways Indicate Active PH Neutral 2,3-Butanedione Fermentation. ISME J. 2014, 8, 1247–1258.

(8)

Serra, R.; Grande, R.; Butrico, L.; Rossi, A.; Settimio, U. F.; Caroleo, B.; Amato, B.; Gallelli, L.; Franciscis, S. de. Chronic Wound Infections: The Role of Pseudomonas Aeruginosa and Staphylococcus Aureus. Expert Rev. Anti Infect. Ther. 2015, 13, 605–613.

(9)

Davies, J. C. Pseudomonas Aeruginosa in Cystic Fibrosis: Pathogenesis and Persistence. Paediatr. Respir. Rev. 2002, 3, 128–134.

ACS Paragon Plus Environment

25

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 38

(10) Lorenz, B.; Wichmann, C.; Stöckel, S.; Rösch, P.; Popp, J. Cultivation-Free Raman Spectroscopic Investigations of Bacteria. Trends Microbiol. 2017, 25, 413–424. (11) Wu, X.; Huang, Y.-W.; Park, B.; Tripp, R. A.; Zhao, Y. Differentiation and Classification of Bacteria Using Vancomycin Functionalized Silver Nanorods Array Based SurfaceEnhanced Raman Spectroscopy and Chemometric Analysis. Talanta 2015, 139, 96–103. (12) Ghebremedhin, M.; Heitkamp, R.; Yesupriya, S.; Clay, B.; Crane, N. J. Accurate and Rapid Differentiation of Acinetobacter Baumannii Strains by Raman Spectroscopy: A Comparative Study. J. Clin. Microbiol. 2017, 55, 2480–2490. (13) Berry, D.; Mader, E.; Lee, T. K.; Woebken, D.; Wang, Y.; Zhu, D.; Palatinszky, M.; Schintlmeister, A.; Schmid, M. C.; Hanson, B. T.; et al. Tracking Heavy Water (D2O) Incorporation for Identifying and Sorting Active Microbial Cells. Proc. Natl. Acad. Sci. 2015, 112, E194–E203. (14) Bowden, J. A.; Heckert, A.; Ulmer, C. Z.; Jones, C. M.; Koelmel, J. P.; Abdullah, L.; Ahonen, L.; Alnouti, Y.; Armando, A.; Asara, J. M.; et al. Harmonizing Lipidomics: NIST Interlaboratory Comparison Exercise for Lipidomics Using Standard Reference Material 1950 Metabolites in Frozen Human Plasma. J. Lipid Res. 2017, 58, 2275–2288. (15) Aksenov, A. A.; Silva, R. da; Knight, R.; Lopes, N. P.; Dorrestein, P. C. Global Chemical Analysis of Biology by Mass Spectrometry. Nat. Rev. Chem. 2017, 1, s41570. (16) Fessenden, M. Metabolomics: Small Molecules, Single Cells. Nature 2016, 540, 540153a. (17) Yu, Z.; Kastenmüller, G.; He, Y.; Belcredi, P.; Möller, G.; Prehn, C.; Mendes, J.; Wahl, S.; Roemisch-Margl, W.; Ceglarek, U.; et al. Differences between Human Plasma and Serum Metabolite Profiles. PLOS ONE 2011, 6, e21230.

ACS Paragon Plus Environment

26

Page 27 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

(18) Keleştemur, S.; Çulha, M. Understanding and Discrimination of Biofilms of Clinically Relevant Microorganisms Using Surface-Enhanced Raman Scattering. Appl. Spectrosc. 2017, 71, 1180–1188. (19) Žukovskaja, O.; Jahn, I. J.; Weber, K.; Cialla-May, D.; Popp, J. Detection of Pseudomonas Aeruginosa Metabolite Pyocyanin in Water and Saliva by Employing the SERS Technique. Sensors 2017, 17, 1704. (20) Bodelón, G.; Montes-García, V.; López-Puente, V.; Hill, E. H.; Hamon, C.; Sanz-Ortiz, M. N.; Rodal-Cedeira, S.; Costas, C.; Celiksoy, S.; Pérez-Juste, I.; et al. Detection and Imaging of Quorum Sensing in Pseudomonas Aeruginosa Biofilm Communities by Surface-Enhanced Resonance Raman Scattering. Nat. Mater. 2016, 15, 1203–1211. (21) Nam, J.-M.; Oh, J.-W.; Lee, H.; Suh, Y. D. Plasmonic Nanogap-Enhanced Raman Scattering with Nanoparticles. Acc. Chem. Res. 2016, 49, 2746–2755. (22) Lee, J.-H.; Nam, J.-M.; Jeon, K.-S.; Lim, D.-K.; Kim, H.; Kwon, S.; Lee, H.; Suh, Y. D. Tuning and Maximizing the Single-Molecule Surface-Enhanced Raman Scattering from DNA-Tethered Nanodumbbells. ACS Nano 2012, 6, 9574–9584. (23) Zhu, W.; Crozier, K. B. Quantum Mechanical Limit to Plasmonic Enhancement as Observed by Surface-Enhanced Raman Scattering. Nat. Commun. 2014, 5, 5228. (24) Le Ru, E. C.; Grand, J.; Sow, I.; Somerville, W. R. C.; Etchegoin, P. G.; TreguerDelapierre, M.; Charron, G.; Félidj, N.; Lévi, G.; Aubard, J. A Scheme for Detecting Every Single Target Molecule with Surface-Enhanced Raman Spectroscopy. Nano Lett. 2011, 11, 5013–5019.

ACS Paragon Plus Environment

27

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 38

(25) Le Ru, E. C.; Meyer, M.; Etchegoin, P. G. Proof of Single-Molecule Sensitivity in Surface Enhanced Raman Scattering (SERS) by Means of a Two-Analyte Technique. J. Phys. Chem. B 2006, 110, 1944–1948. (26) Zrimsek, A. B.; Wong, N. L.; Van Duyne, R. P. Single Molecule Surface-Enhanced Raman Spectroscopy: A Critical Analysis of the Bianalyte versus Isotopologue Proof. J. Phys. Chem. C 2016, 120, 5133–5142. (27) Woehl, T. J.; Heatley, K. L.; Dutcher, C. S.; Talken, N. H.; Ristenpart, W. D. ElectrolyteDependent Aggregation of Colloidal Particles near Electrodes in Oscillatory Electric Fields. Langmuir 2014, 30, 4887–4894. (28) Nadal, F.; Argoul, F.; Hanusse, P.; Pouligny, B.; Ajdari, A. Electrically Induced Interactions between Colloidal Particles in the Vicinity of a Conducting Plane. Phys. Rev. E 2002, 65, 061409. (29) Böhmer. In Situ Observation of 2-Dimensional Clustering during Electrophoretic Deposition. Langmuir 1996, 12, 5747–5750. (30) Sides, P. J. Electrohydrodynamic Particle Aggregation on an Electrode Driven by an Alternating Electric Field Normal to It. Langmuir 2001, 17, 5791–5800. (31) Thrift, W. J.; Nguyen, C. Q.; Darvishzadeh-Varcheie, M.; Zare, S.; Sharac, N.; Sanderson, R. N.; Dupper, T. J.; Hochbaum, A. I.; Capolino, F.; Abdolhosseini Qomi, M. J.; et al. Driving Chemical Reactions in Plasmonic Nanogaps with Electrohydrodynamic Flow. ACS Nano 2017, 11, 11317–11329. (32) Koh, A. L.; Bao, K.; Khan, I.; Smith, W. E.; Kothleitner, G.; Nordlander, P.; Maier, S. A.; McComb, D. W. Electron Energy-Loss Spectroscopy (EELS) of Surface Plasmons in

ACS Paragon Plus Environment

28

Page 29 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Single Silver Nanoparticles and Dimers: Influence of Beam Damage and Mapping of Dark Modes. ACS Nano 2009, 3, 3015–3022. (33) Lyandres, O.; Shah, N. C.; Yonzon, C. R.; Walsh, J. T.; Glucksberg, M. R.; Van Duyne, R. P. Real-Time Glucose Sensing by Surface-Enhanced Raman Spectroscopy in Bovine Plasma Facilitated by a Mixed Decanethiol/Mercaptohexanol Partition Layer. Anal. Chem. 2005, 77, 6134–6139. (34) Cheung, W.; Shadi, I. T.; Xu, Y.; Goodacre, R. Quantitative Analysis of the Banned Food Dye Sudan-1 Using Surface Enhanced Raman Scattering with Multivariate Chemometrics. J. Phys. Chem. C 2010, 114, 7285–7290. (35) Dong, R.; Weng, S.; Yang, L.; Liu, J. Detection and Direct Readout of Drugs in Human Urine Using Dynamic Surface-Enhanced Raman Spectroscopy and Support Vector Machines. Anal. Chem. 2015, 87, 2937–2944. (36) Trau, M.; Saville, D. A.; Aksay, I. A. Assembly of Colloidal Crystals at Electrode Interfaces. Langmuir 1997, 13, 6375–6381. (37) Squires, T. M.; Bazant, M. Z. Induced-Charge Electro-Osmosis. J. Fluid Mech. 2004, 509, 217–252. (38) Watson, D.; MacDERMOT, J.; Wilson, R.; Cole, P. J.; Taylor, G. W. Purification and Structural Analysis of Pyocyanin and 1-Hydroxyphenazine. Eur. J. Biochem. 1986, 159, 309–313. (39) Wu, X.; Chen, J.; Li, X.; Zhao, Y.; Zughaier, S. M. Culture-Free Diagnostics of Pseudomonas Aeruginosa Infection by Silver Nanorod Array Based SERS from Clinical Sputum Samples. Nanomedicine Nanotechnol. Biol. Med. 2014, 10, 1863–1870.

ACS Paragon Plus Environment

29

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 38

(40) Armbruster, D. A.; Pry, T. Limit of Blank, Limit of Detection and Limit of Quantitation. Clin. Biochem. Rev. 2008, 29, S49–S52. (41) Qi, G.; Jia, K.; Fu, C.; Xu, S.; Xu, W. A Highly Sensitive SERS Sensor for Quantitative Analysis of Glucose Based on the Chemical Etching of Silver Nanoparticles. J. Opt. 2015, 17, 114020. (42) Grubisha, D. S.; Lipert, R. J.; Park, H.-Y.; Driskell, J.; Porter, M. D. Femtomolar Detection of Prostate-Specific Antigen:  An Immunoassay Based on Surface-Enhanced Raman Scattering and Immunogold Labels. Anal. Chem. 2003, 75, 5936–5943. (43) Srichan, C.; Ekpanyapong, M.; Horprathum, M.; Eiamchai, P.; Nuntawong, N.; Phokharatkul, D.; Danvirutai, P.; Bohez, E.; Wisitsoraat, A.; Tuantranont, A. HighlySensitive Surface-Enhanced Raman Spectroscopy (SERS)-Based Chemical Sensor Using 3D Graphene Foam Decorated with Silver Nanoparticles as SERS Substrate. Sci. Rep. 2016, 6, 23733. (44) Lee, D.; Lee, S.; Seong, G. H.; Choo, J.; Lee, E. K.; Gweon, D.-G.; Lee, S. Quantitative Analysis of Methyl Parathion Pesticides in a Polydimethylsiloxane Microfluidic Channel Using Confocal Surface-Enhanced Raman Spectroscopy. Appl. Spectrosc. 2006, 60, 373– 377. (45) Stevenson, R.; Ingram, A.; Leung, H.; C. McMillan, D.; Graham, D. Quantitative SERRS Immunoassay for the Detection of Human PSA. Analyst 2009, 134, 842–844. (46) Theis, K. R.; Venkataraman, A.; Dycus, J. A.; Koonter, K. D.; Schmitt-Matzen, E. N.; Wagner, A. P.; Holekamp, K. E.; Schmidt, T. M. Symbiotic Bacteria Appear to Mediate Hyena Social Odors. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 19832–19837.

ACS Paragon Plus Environment

30

Page 31 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

(47) Nguyen, C.; Thrift, W.; Bhattacharjee, A.; Whiteson, K.; Hochbaum, A.; Ragan, R. Robust SERS Spectral Analysis for Quantitative Detection of Pycocyanin in Biological Fluids. In Biosensing and Nanomedicine X, Proceedings of SPIE Nanoscience + Engineering, San Diego, CA, Aug 6–10, 2017; International Society for Optics and Photonics, 2017. (48) Reszka, K. J.; O’Malley, Y.; McCormick, M. L.; Denning, G. M.; Britigan, B. E. Oxidation of Pyocyanin, a Cytotoxic Product from Pseudomonas Aeruginosa, by Microperoxidase 11 and Hydrogen Peroxide. Free Radic. Biol. Med. 2004, 36, 1448– 1459. (49) Dietrich, L. E. P.; Price-Whelan, A.; Petersen, A.; Whiteley, M.; Newman, D. K. The Phenazine Pyocyanin Is a Terminal Signalling Factor in the Quorum Sensing Network of Pseudomonas Aeruginosa. Mol. Microbiol. 2006, 61, 1308–1321. (50) Bhattacharjee, A.; Khan, M.; Kleiman, M.; Hochbaum, A. I. Effects of Growth Surface Topography on Bacterial Signaling in Coculture Biofilms. ACS Appl. Mater. Interfaces 2017, 9, 18531–18539. (51) Adams, S. M.; Campione, S.; Caldwell, J. D.; Bezares, F. J.; Culbertson, J. C.; Capolino, F.; Ragan, R. Non-Lithographic SERS Substrates: Tailoring Surface Chemistry for Au Nanoparticle Cluster Assembly. Small 2012, 8, 2239–2249. (52) Rahme, L. G.; Stevens, E. J.; Wolfort, S. F.; Shao, J.; Tompkins, R. G.; Ausubel, F. M. Common Virulence Factors for Bacterial Pathogenicity in Plants and Animals. Science 1995, 268, 1899–1902. (53) Monahan, L. G.; Turnbull, L.; Osvath, S. R.; Birch, D.; Charles, I. G.; Whitchurch, C. B. Rapid Conversion of Pseudomonas Aeruginosa to a Spherical Cell Morphotype Facilitates

ACS Paragon Plus Environment

31

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 38

Tolerance to Carbapenems and Penicillins but Increases Susceptibility to Antimicrobial Peptides. Antimicrob. Agents Chemother. 2014, 58, 1956–1962. (54) Parker, J. H.; Feldman, D. W.; Ashkin, M. Raman Scattering by Silicon and Germanium. Phys. Rev. 1967, 155, 712–714. (55) Geladi, P.; Kowalski, B. R. Partial Least-Squares Regression: A Tutorial. Anal. Chim. Acta 1986, 185, 1–17. (56) Adams, S. M.; Campione, S.; Capolino, F.; Ragan, R. Directing Cluster Formation of Au Nanoparticles from Colloidal Solution. Langmuir 2013, 29, 4242–4251. (57) Grady, N. K.; Halas, N. J.; Nordlander, P. Influence of Dielectric Function Properties on the Optical Response of Plasmon Resonant Metallic Nanoparticles. Chem. Phys. Lett. 2004, 399, 167–171. (58) Moerland, R. J.; Hoogenboom, J. P. Subnanometer-Accuracy Optical Distance Ruler Based on Fluorescence Quenching by Transparent Conductors. Optica 2016, 3, 112–117.

ACS Paragon Plus Environment

32

Page 33 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Figure 1. Schematic overview of chemical assembly of SERS substrates. (a) Lipoic acid-functionalized Au nanoparticles are electrophoretically driven toward a working electrode. Au seeds chemically crosslink with amine functionalized PMMA regions on self-organized diblock copolymer PS-b-PMMA template (inset). The resultant EHD flow field around the seed entrains nearby nanoparticles forming transient oligomers. (b) Illustration of anhydride bridge that forms between carboxylic acid and acylisourea on neighboring nanoparticles that stabilizes oligomers when the external field is turned off. 41x20mm (300 x 300 DPI)

ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. (a) SEM micrograph of self-assembled Au oligomers. (b) Attenuation curve of SERS substrates in DI water. Dotted curves represent absorption cross section of a dimer (×), linear trimer (squares), and linear quadrumer (+) from full-wave simulations. Inset shows oligomers distributions in percent area on SERS substrate calculated from SEM images acquired over 10 µm2. (c) Calculated field enhancement of Au (i) linear trimer with 0.9 nm gap, (ii) dimer with 0.9 nm gap, (iii) linear trimer with 2 nm gap, and (iv) dimer with 2 nm gap from full-wave simulations. Curve (i) and (ii) are offset by 100 for clarity. (d) Normalized SERS intensity map of benzenethiol’s 1573 cm-1 vibrational band. Inset shows distribution of normalized intensity with a RSD of 10.4%. 177x128mm (286 x 286 DPI)

ACS Paragon Plus Environment

Page 34 of 38

Page 35 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Figure 3. (a) SERRS spectra of (i) DI water, (ii) 1 µg·mL-1, (iii) 10 µg·mL-1, and (iv) 100 µg·mL-1 pyocyanin in water. Grey bars indicate Raman bands of pyocyanin. (b) SERRS spectra of of (i) DI water, (ii) 100 pg·mL1 , (iii) 1 ng·mL-1, (iv) 10 ng·mL-1, and (v) 100 ng·mL-1 pyocyanin in water. Dark grey bar highlights pyocyanin’s 552 cm-1 band while light grey bars indicate other pyocyanin Raman bands. (c) Dose-response relationship of pyocyanin and normalized SERS intensity at 552 cm-1. Error bars depict standard deviation calculated from 100 measurements. 54x16mm (300 x 300 DPI)

ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. (a) SERRS fingerprint of (i) bacteria-free supernatant from mid-stationary phase cultures of wildtype P. aeruginosa PA14, (ii) 10 µg·mL-1 pyocyanin in LB broth, (iii) phenzine-null mutant strain ∆phz1/2 and (iv) LB broth. (b) Pyocyanin concentration predicted by PLS model for spectra in testing set. Error bars depict standard deviation calculated from 20 measurements. Inset shows RMSECV of the PLS model with respect to the number of components used. (c) Pyocyanin concentration in bacteria-free supernatants of wild-type P. aeruginosa PA14 and its phenazine-null strain over 24 hours. Concentrations from SERS and UV-Vis measurements are calculated using trained PLS model and reported molar absorptivity ε = 4.31 × 103 mol-1 cm-1 at λmax = 690 nm (from reference 48), respectively. Error bars show standard deviation calculated from 400 measurements. (d) Growth curves for planktonic cultures of P. aeruginosa PA14 and its phenazine-null mutant strain as determined from optical density. 178x164mm (231 x 231 DPI)

ACS Paragon Plus Environment

Page 36 of 38

Page 37 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Figure 5. (a) Schematic of the biofilm in-line measurements setup. (b) Pyocyanin concentration predicted by PLS model from SERS spectra of various time. Each data point represents the averaged of predicted concentrations and error bars show standard deviation calculated from 200 measurements. (c) Individual representative SERS spectra at (i) 12h (ii) 9h, (iii) 6h, and (iv) 3h contrasted with spectrum of (v) TB media. Grey bars indicate pyocyanin vibrational bands. (d) Accumulated biomass of biofilm growth in the flow cell over time. Error bars depict standard deviations calculated from n = 3 independent growth channels. Confocal fluorescence microscopy images show representative bacterial accumulation on the glass surface at the time points indicated by the arrows (scale bars are 20 µm). (e) Susceptibility of surfaceattached bacteria to carbenicillin treatment at the specified growth times. Error bars depict standard deviation of dead cell fraction calculated from n = 3 biological replicates. 118x79mm (300 x 300 DPI)

ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table of Contents Graphics 35x15mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 38 of 38